The usability of color and near infrared reflection data in determination of adulteration in dried and powdered organic black carrot

Adulteration is a significant threat especially in powdered food products. The drying method, one of the stages before powdering, can cause different quality changes on the product. Therefore, the drying method may affect the detection success of adulteration. In this study, the discernibility of the organic and same cultivar conventional black carrots, dried by hot air (HA) and intermittent microwave (IMW), were investigated with color (L*, a*, b*) and FT-NIRS (1000-2500 nm) sysytems. In addition, the adulteration level (%) estimation and distinguishability of conventional powder samples and adulterated organic powder samples with these systems were investigated by PCA (Principal Component Analysis), SIMCA (Soft Independent Modeling Class Analogy) and PLSR (Partial Least Square Regression) methods. Also, the IMW and HA drying kinetics of the samples were examined and the experimental drying kinetics data were modeled mathematically. Also, fresh and dried samples were compared in terms of total phenol and total antioxidant. The variation of mean L* and b* values of fresh organic and conventional samples was significantly different (p<0.001). The PCA results of the color and FT-NIRS data of the pure powder samples were obtained a total explained variability of over 90% with first two PCs. Results of SIMCA, showed that 100% (6/6*100) and 50% (3/6*100) of the pure organic samples color data dried with HA and IMW, respectively could be classified correctly. By using FTNIRS data, it was determined that the samples could not be classified as organic and conventional with the SIMCA method. The best models for estimating the level of adulteration (%) as a result of PLSR were obtained with FT-NIRS data (R2val=0.98, RMSEP=4.9% and R2val=0.89, RMSEP=10.1% for HA and IMW, respectively). Models with values R2val=0.82, RMSEP=13.2% and R2val=0.53, RMSEP=21.2% values, respectively, were obtained for HA and IMW as a result of PLSR of color data. The samples dried by using HA were differentiated relatively better than IMW, and FT-NIRS data were more reliable tool compared to colors data in determining the level of adulteration (%). This may be related to running of FT-NIRS in a wider spectral bandwidth. From the results of drying kinetic analysis, it was determined that the best fit models were the Verma model for HA and the Logistic model for IMW. Total phenol and total antioxidant capacity values were found to be higher in fresh organic samples. The total phenol and total antioxidant quantities of the samples dried by applying 300 and 450 W power levels via IMW were higher than dried by HA. Moreover 450 W was the best power level in terms of protecting and increasing the total phenol and antioxidant capacity level.